Targeting early changes in the synovial microenvironment: a new class of immunomodulatory therapy?

Abstract

Objectives: Controlled immune responses rely on integrated crosstalk between cells and their microenvironment. We investigated whether targeting proinflammatory signals from the extracellular matrix that persist during pathological inflammation provides a viable strategy to treat rheumatoid arthritis (RA).

Methods: Monoclonal antibodies recognising the fibrinogen-like globe (FBG) of tenascin-C were generated by phage display. Clones that neutralised FBG activation of toll-like receptor 4 (TLR4), without impacting pathogenic TLR4 activation, were epitope mapped by crystallography. Antibodies stained synovial biopsies of patients at different stages of RA development. Antibody efficacy in preventing RA synovial cell cytokine release, and in modulating collagen-induced arthritis in rats, was assessed.

Results: Tenascin-C is expressed early in the development of RA, even before disease diagnosis, with higher levels in the joints of people with synovitis who eventually developed RA than in people whose synovitis spontaneously resolved. Anti-FBG antibodies inhibited cytokine release by RA synovial cells and prevented disease progression and tissue destruction during collagen-induced arthritis.

Conclusions: Early changes in the synovial microenvironment contribute to RA progression; blocking proinflammatory signals from the matrix can ameliorate experimental arthritis. These data highlight a new drug class that could offer early, disease-specific immune modulation in RA, without engendering global immune suppression.

Keywords: extracellular matrix; inflammation; monoclonal antibodies; rheumatoid arthritis; tenascin-C.

Figure 1

Generation and affinity maturation of anti-tenascin-C antibodies. (A) Fab clones were screened for their ability to inhibit FBG-mediated upregulation of nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) in THP1-blue cells. (B) The potency of top hits (clones 15, 17 and 20) was further assessed compared with control Fab D1.3 recognising hen egg lysozyme. (A and B) Data are shown as the mean±SD from three experiments. (C) The binding affinity of NSC20 to human FBG was analysed by surface plasmon resonance (SPR) spectroscopy at 25°C. (D) Following affinity maturation, NSC20 variants were screened for improved binding to human and mouse FBG. HTRF measured the ability of affinity matured scFv clones to compete with labelled NCS20 IgG for antigen binding. Data are presented as percentage inhibition of the FRET signal (x-axis). The ENC assay used limiting amounts of immobilised anti-FLAG antibody to capture FLAG-tagged scFv from culture supernatants and thereby normalise the amount of immobilised scFv. After washing to remove excess (unbound) scFv, the ability of immobilised antibody clones to bind biotinylated CD4-his-mTNC-FBG was detected by DELFIA, allowing ranking of clones based on affinity. Parental NSC20 was used as a benchmark, and data are presented as fold-increase in fluorescence signal (y-axis). Open circles represent NSC20 variants, solid green circle indicates lead variant, C3. (E) The heavy chain CDR3 (HCDR3) sequence and binding properties, determined by SPR at 37°C, of C3 are shown compared with the parental antibody NSC20. In contrast to NSC20, no measurable dissociation of the antibody–antigen complex was detectable over a 10 min period at 25°C for C3 (data not shown). Hence, affinity values for C3 were determined at 37°C. ENC, expression-normalised capture; FBG, fibrinogen-like globe; HTRF, homogeneous time resolved fluorescence.

Figure 2

Defining the mode of action of neutralising anti-tenascin-C antibodies. (A and B) The crystal structure of C3 bound to the FBG domain of human tenascin-C. Six FBG residues bind to Fab chain B (heavy chain) (A2115, Y2116, R2147, R2151, N2118 and H2171) and 13 FBG residues with Fab chain C (light chain) (Y2116, S2131, I2133, Y2140, R2147, N2148, C2149, H2150, R2151, H2163, S2164, F2170 and H2175), three of which are shared (Y2116, R2147 and R2151). Light green: FBG residues in contact with Fab chain B; dark green: FBG residues in contact with Fab chain C; blue: FBG residues in contact with Fab chains B and C. This interaction interface is not conserved in the FBG domains of tenascin-R and tenascin–W. Only 9 of these 16 residues are present in FBG-R, and only seven in FBG-W. Of the three residues in FBG-C that interact with both heavy and light chain of the antibody, one residue is substituted in FBG-R, and all three are substituted in FBG-W. These data support experimental evidence that FBG-R does not bind to C3, and the higher sequence divergence of the FBG domain of tenascin-W with that of tenascin-C (54.1%) compared with tenascin-R (61.6%), and the fact that this divergence includes key positions in the C3 binding epitope, predicts that FBG-W will also be unable to bind to C3. (C) Recombinant human TLR4 was coated onto a 96-well plate, and recombinant human tenascin-C FBG, which had been preincubated with C3 or isotype control antibody, was added. Bound FBG was detected, and the percentage inhibition in the C3 preincubated samples was calculated compared with the isotype control samples (IC50=45.5 nM). Data are shown as the mean±SEM from eight experiments. (D) Recombinant human tenascin-C FBG (FBG) (1 µM) or lipopolysaccharide (LPS) (1 ng/mL) were preincubated with the indicated doses of C3 or isotype control antibody (Ig) before being added to primary human macrophages. After 24 hours, supernatants were taken, and cytokine ELISAs were performed. Data are shown as the mean±SEM from four independent donors. One-way analysis of variance was performed to determine significance of C3 inhibition compared with isotype control. *p<0.05, **p<0.01, ****p<0.0001. FBG, fibrinogen-like globe; IL, interleukin; TLR4, toll-like receptor 4; TNF, tumour necrosis factor.

Figure 3

Antibody treatment ameliorates inflammation in experimental arthritis. (A and B) Synovial biopsies from the Birmingham early arthritis cohort were stained with anti-FBG antibodies. Quantification of the number of positive pixels per microgram biopsy was performed in tissue from people with early RA (undiagnosed synovitis of less than 3 months, patients who go on to be diagnosed with RA), people with synovial inflammation that spontaneously resolved (undiagnosed synovitis of less than 3 months that disappears by itself) (resolving) and people with established RA (diagnosed disease, greater than 3 months’ duration) (A). Anti-FBG staining was observed in areas of fibrosis in inflamed synovia (left panel: red anti-FBG, green anti-CD90, blue antipodoplanin) and around areas of vascularisation underneath endothelial cells (centre panel: red anti-FBG, green anti-CD31). No staining was observed with isotype controls in place of primary antibodies (right panel). Scale bars 50 µm (B). (C) Mixed populations of cells isolated from RA synovial membranes were stimulated with 1 µM recombinant human tenascin-C FBG or 1 ng/mL LPS, which had been preincubated with either C3 or isotype control antibody. After 24 hours, supernatants were taken and cytokine ELISAs were performed. Data are shown as the mean±SEM from three independent donors. One-way analysis of variance (ANOVA) was performed to determine significance of C3 inhibition compared with isotype control. *p=0.02, ***p<0.0001. (D–F) Rats were injected with bovine type II collagen intradermally on day 0 and day 7. Treatments of PBS (vehicle control) and C3 at 1, 3 or 10 mg/kg were administered twice weekly throughout the experiment by intravenous injection (10 animals per treatment group). Animals were scored for clinical signs of disease three times per week, and the mean±SEM is shown (D). Paw volumes were measured using a plethysmometer on days 0, 14, 21 and 28, and the mean change±SEM normalised to the day 0 measurements is shown (E). Two-way was carried out to test for significance of changes between vehicle and treatment groups. ****p<0.0001, ***p<0.001, **p<0.01, *p<0.05. (F) At termination on day 28, hind limbs were assessed for histological signs of inflammation, articular cartilage damage and damage to the underlying metaphyseal bone. A χ² test for trend was used to confirm that the presence or absence of histopathological signs is associated with antibody treatment (χ²=9.098, p=0.003), with only 3 of 20 paws were free of any sign of histopathology in the vehicle treated group, whereas 11 of 20 paws were disease free in the 10 mg/kg treated group (table). Representative histological images of destructive arthritis (vehicle treated) and a normal joint (10 mg/kg) are shown (right panels). FBG, fibrinogen-like globe; IL, interleukin; RA, rheumatoid arthritis; TNF, tumour necrosis factor.